info: adapt deduplicated archive size calculation to match borg create stats

also: test that both info and create output the same "this archive" "deduplicated size" stats.

Author: ThomasWaldmann

src/borg/archive.py

@@ -681,21 +681,23 @@ def _calc_stats(self, cache, want_unique=True):
         if have_borg12_meta and not want_unique:
             unique_csize = 0
         else:
-            def add(id):
-                entry = cache.chunks[id]
-                archive_index.add(id, 1, entry.size, entry.csize)
 
             archive_index = ChunkIndex()
             sync = CacheSynchronizer(archive_index)
-            add(self.id)
+            # do NOT add the archive metadata chunk (self.id) here.
+            # The metadata chunk is accounted via meta_stats during creation and must not
+            # contribute to the "This archive" deduplicated size computed by borg info.
+            # See issue #9003: make info's deduplicated size match create-time stats.
+
             # we must escape any % char in the archive name, because we use it in a format string, see #6500
             arch_name_escd = self.name.replace('%', '%%')
             pi = ProgressIndicatorPercent(total=len(self.metadata.items),
                                           msg='Calculating statistics for archive %s ... %%3.0f%%%%' % arch_name_escd,
                                           msgid='archive.calc_stats')
             for id, chunk in zip(self.metadata.items, self.repository.get_many(self.metadata.items)):
                 pi.show(increase=1)
-                add(id)
+                # do NOT add(id) here, this is a metadata stream chunk and should not
+                # be accounted for in stats, see comment above.
                 data = self.key.decrypt(id, chunk)
                 sync.feed(data)
             unique_csize = archive_index.stats_against(cache.chunks)[3]

src/borg/testsuite/archiver.py

@@ -1617,6 +1617,27 @@ def test_info(self):
         info_archive = self.cmd('info', '--first', '1', self.repository_location)
         assert 'Archive name: test\n' in info_archive
 
+    def test_info_matches_create_deduplicated_size(self):
+        # Create two identical files to ensure intra-archive deduplication happens,
+        # so the deduplicated size is visibly different from compressed size.
+        data = b'X' * (1024 * 80)
+        self.create_regular_file('file1', contents=data)
+        self.create_regular_file('file2', contents=data)
+        self.cmd('init', '--encryption=repokey', self.repository_location)
+        create_out = self.cmd('create', '--stats', self.repository_location + '::test', 'input')
+        info_out = self.cmd('info', self.repository_location + '::test')
+
+        import re
+
+        def get_dedup(s):
+            m = re.search(r'^This archive:\s+(.*?)\s+(.*?)\s+(.+)$', s, re.M)
+            assert m is not None, s
+            return m.group(3).strip()
+
+        dedup_create = get_dedup(create_out)
+        dedup_info = get_dedup(info_out)
+        assert dedup_create == dedup_info
+
     def test_info_json(self):
         self.create_regular_file('file1', size=1024 * 80)
         self.cmd('init', '--encryption=repokey', self.repository_location)